Identifying author heritage using surname data: An application for Russian surnames

Abstract

This research article puts forward a method to identify the national heritage of authors based on the morphology of their surnames. Most studies in the field use variants of dictionary‐based surname methods to identify ethnic communities, an approach that suffers from methodological limitations. Using the public file of ORCID (Open Researcher and Contributor ID) identifiers in 2015, we developed a surname‐based identification method and applied it to infer Russian heritage from suffix‐based morphological regularities. The method was developed conceptually and tested in an undersampled control set. Identification based on surname morphology was then complemented by using first‐name data to eliminate false‐positive results. The method achieved 98% precision and 94% recall rates—superior to most other methods that use name data. The procedure can be adapted to identify the heritage of a variety of national groups with morphologically regular naming traditions. We elaborate on how the method can be employed to overcome long‐standing limitations of using name data in bibliometric datasets. This identification method can contribute to advancing research in scientific mobility and migration, patenting by certain groups, publishing and collaboration, transnational and scientific diaspora links, and the effects of diversity on the innovative performance of organizations, regions, and countries.

'Moving On' -- Investigating Inventors' Ethnic Origins Using Supervised Learning

“…Ferrucci, 2020;Akcigit et al, 2016;Ghani et al, 2014;Gaule and Piacentini, 2013). But there are some important challenges related to such name-matching techniques (see Karaulova et al 2019or Breschi et al 2017 for an overview). First of all, name matching approaches generally have a low recall, which means that they tend to miss a substantial fraction of inventors from a given ethnic origin (Karaulova et al, 2019).…”

“…This is a relatively high performance with regard to related approaches in the literature that report performance scores (see e.g. Karaulova et al, 2019;Ye et al, 2017). Using this trained classification model, I predict and investigate the ethnic origins of over 2.68 million inventors from patents filed at the European Patent Office (EPO) or the United States Patent and Trademark Office (USPTO) between 1980 and 2015.…”

“…My paper primarily differs from these contributions with regard to their methodological approaches as well as the scopes of their analyses, which are often restricted to one specific country or ethnic origin. For example, Karaulova et al (2019) present a stepwise procedure that is primarily based on the morphology of surnames. Their proposed approach does not rely on a commercial database, but they focus exclusively on Russian origin.…”

Valuation regimes in academia: Researchers’ attitudes towards their diversity of activities and academic performance

“…Finally, ethnicity has been found to be a feature related with collaboration; those from the same ethnicities will be more likely to work together (Freeman and Huang 2014). Ethnicity has been used also as a proxy to identify academics with a common cultural background (Karaulova et al 2019).…”

Predicting perceived ethnicity with data on personal names in Russia

“…The most well known reference list of Russian surnames compiled by Unbegaun (1972) is incomplete and does not directly link surnames to ethnic groups. Karaulova et al (2019) develop a method for identifying ethnically Russian surnames that uses suffix-based morphological regularities. ML based methods can achieve higher accuracy, and, besides, the method proposed by Karaulova et al (2019) cannot distinguish between various ethnic groups populating Russia that is home to over 100 ethnic groups with often characteristic personal names.…”